dim msx basic | MSX Resource Center (ページ 2/3)

Champion (301)

02-10-2017, 23:13

Maybe a simple trick is not to use a 2d array but just a long '1D', that might remove 'some' of the multiplications for the look-up if your algorithm allows to incrementally traverse the array somehow.

By DamnedAngel

Champion (266)

02-10-2017, 23:40

@TheKid,
You mentioned 256 values, but you meant 512, right (32x16)?

@All,
I just made a experience in WEBMSX to compare DIM with POKE times (supposing each TheKid's data can be accommodated in bytes), and the results were not impressive:

List:

Edit: The image didn't show up, so here ir the URL: https://msx.pics/image/D77r

Run:

Edit: The image didn't show up, so here ir the URL: https://msx.pics/image/DnjC

Didn't everyone expect POKE to be somewhat faster? DId I do some dumb thing in my basic code?

I don't know how time accurate WEBMSX is, but since it plays games very well I think its results should be representative.

By wolf_

Ambassador_ (10109)

03-10-2017, 00:23

Well, "poke 12345,255" is slightly longer to parse than "a(12345)=255". Maybe that's a factor too?

By TheKid

Paragon (1238)

03-10-2017, 07:37

@DamnedAngel: You are right, 32x16 isn't 256 But nice test, I was under the impression that using pokes would be much faster dan dim.

By Lord_Zett

Paladin (807)

03-10-2017, 10:10

eeej The Kid. using poke is easy in basic. its faster and you can bsave the data.

do:

Clear XXXX, &HBFFF

The XXXX is the size. max 4096 bytes to use. but i do not use more than 2048 bytes
Then you can poke &HC000+(byte nummer),data

to read D=poke(&Hc000+(byte nummer))

By TheKid

Paragon (1238)

03-10-2017, 10:13

@Thanks Lord_Zett, I will give it a try and hope it's faster then the dim array.

By Lord_Zett

Paladin (807)

03-10-2017, 10:18

yeah the easy bsave and bload the data is handy. you can just bsave "data.dat",&HC000, &HC000+(byte nummer max)

By Arjan

Paladin (787)

03-10-2017, 11:05

@DamnedAngel you defined variables I, J and X as INT, but you didn't define variable K, that's why the POKE version is not as fast as expected. If K is also INT, the POKE version is almost twice as fast!

By NYYRIKKI

Enlighted (6067)

03-10-2017, 12:48

I suggest that you make separate program that reads the text file. Then in the end you use COPY-command to store the table in to file and in your main program you use COPY again to restore the table values.

See details here.

By DamnedAngel

Champion (266)

03-10-2017, 14:55

Lord_Zett wrote:

to read D=poke(&Hc000+(byte nummer))

To read, use peek instead of poke.

Arjan wrote:

@DamnedAngel you defined variables I, J and X as INT, but you didn't define variable K, that's why the POKE version is not as fast as expected. If K is also INT, the POKE version is almost twice as fast!

It was intentional, since my stupid mind considered that int ranges from -7fffh to 8000h, leaving c000h out of the interval. But, of course c000h = -4000h(16 bits) which is or course within the range. Fool me.

@TheKid, new version, twice as fast, thanks to Arjan:

list:

https://msx.pics/image/DDzQ

run:

https://msx.pics/image/Db5W

[]s